A study on plant recognition based on leaf image visualization systems

Accurate plant recognition is essential for biodiversity, ecological studies, and even the future of agriculture. Traditional methods are often slow and labor-intensive and rely heavily on experts. That’s why automated solutions have become so important. This survey looks at how artificial intelligence is changing the game for plant leaf recognition. It’s all about deep learning, machine learning, and image processing. Convolutional neural networks—think VGG16, ResNet50, DenseNet201 are some examples of models used in this field. Ensemble models and generative approaches add another layer, making systems more robust by expanding data diversity. Preprocessing steps like edge detection, morphological operations, background removal, and even generating synthetic data give the models a real boost in accuracy. For feature extraction, researchers lean on texture, shape, and color descriptors, while explainable AI tools like Grad-CAM make it easier to understand why models make certain decisions. This survey gathers the latest methods, tracks how they perform on different datasets, and points out both the challenges and the big opportunities ahead for building plant recognition systems that are efficient, accurate, and scalable. • Comprehensive review of machine learning and deep learning methods for plant recognition using leaf images. • Comparative analysis of preprocessing, feature extraction, and classification techniques used in recent studies. • Systematic overview of widely used public leaf image datasets and evaluation metrics. • Identification of key challenges, research gaps, and limitations in existing plant recognition systems. • Discussion of future research directions for accurate, interpretable, and deployable leaf-based identification models.